I lost both of my sons. Kevin took his own life. Chris died of drugs. I am building something so that other people don’t have to write a sentence like that one.
This is not a picture. It is the real world we built, at their real coordinates, and you can walk up to them right now without an account. Kevin’s stone has a video game controller on it. Christopher’s has a pencil across an open sketchbook.
Kevin came into the world in a spring blizzard. He was the protective one — everybody who knew him says it, and his brother’s obituary says it too. Chris arrived four days before Christmas, and Kevin used to say he was the best Christmas present he ever got. Chris was the funny one. He drew constantly. He wanted to be a cartoon artist and an animator.
I’m not going to describe how either of them died, and I want you to notice that, because it isn’t squeamishness. There is solid research showing that detail in this kind of story raises the risk for somebody reading it who is already close to the edge. So this is written to the guidelines that protect that person. If you take one thing from it: the way we talk about this is itself a mechanism. It can hurt people or it can help them. Almost nobody treats it as a design decision. I do.
For a long time I thought I had lost my sons to two separate things. One boy to despair, one to drugs. Two tragedies, unrelated except by the house they grew up in.
I don’t think that any more.
That is what took Chris. Counterfeit pills don’t come from a stranger on a corner any more — they come through a phone, from a handle, to a kid. Teen drug use has been flat or falling for years while overdose deaths among ten-to-nineteen-year-olds more than doubled. Fewer children using. Far more children dying. The behaviour didn’t get worse; the supply got lethal and anonymous.
And it is the same shape that reached Kevin. Not pills — a worldview. The most-shared thing a struggling teenage boy encounters online is that nothing matters, delivered as a joke, stripped of every consequence. In the show it came from, that nihilism is a portrait of a man being destroyed by it. In a twenty-second clip on a phone it is a slogan with a laugh track. The framing that makes it a tragedy only survives at full length, and no child watches at full length.
We know this isn’t a feeling. When researchers measured a series that depicted suicide without following those guidelines, youth suicide rose almost 29% in the month after release — and there was no measurable effect on adults at all. Same content. Same country. Same month. Harmful to children, nothing to grown-ups.
That is the whole argument, and it has already been proven. The problem was never only what gets made. It is a delivery system that cannot tell a child from an adult, and was never asked to.
I went looking for what the evidence says works, and it is not what I expected. It is not monitoring. It is not filtering. It is not talking to them more about drugs.
Children are protected by being genuinely known by adults who are not grading them, and by having somewhere they are missed when they don’t turn up. That’s it. That’s the finding, over and over.
There’s a second one almost nobody has heard of. Everyone knows that stories about suicide can increase it. Its opposite is just as real and just as measured: stories about people who reach the edge and find another way through reduce it. Researchers call it the Papageno effect.
Which means the danger isn’t only the despair a child is served. It’s that the despair is everywhere and the way out is nowhere.
Not a safer feed. A different kind of software, with different physics. Every one of these is a mechanism, not a promise — I have tried to be exact about what exists today and what doesn’t.
Withdrawal comes before crisis, and it shows up in attendance long before it shows up in anything anyone writes. A person stops coming to the Thursday night. Drops out of their crew. Goes quiet in rooms they used to be in. Our platform is built on real occasions in real rooms, so it can see that — where a feed only ever has the messages.
When it sees it, it does not send that person a wellbeing pop-up. It quietly tells somebody in their crew: haven’t seen Sam in a couple of weeks. A hotline banner does not save anybody. A specific human being noticing does.
And it is measured against that person’s own baseline, never an average — because solitude and isolation look identical in a snapshot, and only direction tells them apart.
Standing is earned in a real place, in front of witnesses, attached to a real person. Discovery is a vouch, and a vouch is attributable. An unknown handle can’t reach a kid here — not because we filter well, but because the graph has no edge for it. That is the door Chris’s pills came through, closed by construction rather than by moderation.
Clout is zero-sum: for one person to rise, others must fall, so almost everyone loses and they all know it. Depth of connection isn’t — me deepening a friendship costs you nothing. You cannot farm strangers here, so no child has to perform for people who don’t know them.
Loneliness isn’t an absence of people. It’s the absence of anyone who would notice you gone. So we made showing up count: a weekly slot, a crew, a room where the audience has standing too — you don’t have to be good at anything to matter to the people in it. For a boy convinced nobody would notice, that is an argument made by architecture instead of by someone telling him he’s loved, which he will not believe.
People are banned for growth, not for life — there is a route back and it is a public one. If you only protect one kind of child you have built a bunker, and somebody else’s son still dies at the other end of it.
The naive version fails on its own evidence: a dashboard that scores a child is the same achievement pressure that harms them, in a new wrapper. So a parent sees context — which venues, which hosts, which crews, who vouched for them — and patterns, never transcripts. The child sees exactly what the parent sees, always. You get the early warning and the child keeps their privacy, which almost nobody else can offer, because almost nobody else is anchored to real rooms.
The Papageno effect, made into a platform standard rather than an accident. Safe-messaging rules we publish and hold ourselves to — never the method, never the romance, always the resources — and real people telling the story of the part after the worst night.
Not a filter and not a ban. A refusal to let a recommendation engine hand a thirteen-year-old adult content with the consequences edited out, on autoplay, because it has no idea how old he is and no reason to care. Nobody chose that for my sons. An algorithm did, and no one signed it.
I am not going to tell you this would have saved Kevin. I don’t know that, and anybody who tells you they can engineer their way to certainty about a particular child is selling something.
I’m also not telling you a television show killed my son, or that a phone killed my other one. Grief looks for one thing to hold responsible and there is never one thing.
What I am telling you is narrower and I think it is true: the systems that shaped my sons’ last years were built by people who never had to consider them, and they could have been built differently. Every mechanism above is a choice somebody made the other way, for reasons that were mostly commercial and never malicious.
My sons are buried next to each other in a cemetery I have only managed to visit once. I built them a place in our world, at their real coordinates, where anyone can find them. I was asked whether I wanted it kept private.
Kevin was sixteen and he looked after his brother. Chris was seventeen and he drew, and he made everyone laugh, and he wanted to make cartoons.
They should be here. They aren’t. So I am going to spend the rest of my life on the thing that might mean somebody else’s child is.